.jpeg?noresize)
Most UK businesses are now overpaying for cloud, and it took them a decade to realise it. The cloud-first instinct that made sense in 2014 has, by 2026, produced a generation of estates where the bill is opaque, running costs are still hard to forecast and the workloads sitting in the wrong place are eating the innovation budget.
Flexera's 2026 State of the Cloud Report puts UK cloud waste at 29%, the first rise in five years. Across the FinOps engagements my team runs, we typically find 18% to 35% of total cloud spend is recoverable, and nearly a third of Microsoft 365 licences are underused. This isn't a wasteful industry. It's a sector that grew its cloud estate organically, through cloud-first projects, SaaS adoption, departmental purchasing and a recurring cycle of modernisation initiatives, never through a single long-term operational strategy. The current IT budget and cloud bill reflects what that history looks like. Set that alongside recent rises in hardware and software costs for on-premises systems and the optimisation game is really on.
The conversation has shifted accordingly. The dominant assumption used to be that moving workloads to the cloud would improve efficiency and reduce operational complexity. It often did. It also often didn't. By 2026, the question worth asking isn't whether to be in the cloud. It's whether the workloads that moved are still in the right place and indeed whether on-premises workloads still make sense under rising cost pressures.
The Right Workload, the Right Platform
Cost matters more now than it did, and not because the spend itself has changed. Boardrooms have changed. A slow economy, margin pressure, AI investment competing for the same budget pool, and operational resilience expectations have all moved cost from an IT team concern to a CFO and board concern. The strongest cloud strategies now build cost visibility and workload governance into the operating model from the outset, rather than treating them as a reactive exercise after IT spend has started to raise eyebrows.
The savings on the table are significant, and they come from three commercial levers rather than one. Placement matters: cloud-native applications that predominantly run on PaaS and SaaS services earn their keep, while steady-state VM workloads can often run 20% to 40% cheaper on sovereign infrastructure because cloud pricing models do not reward always-on “legacy” workloads. Modernisation matters: applications running on cloud IaaS often belong on cloud PaaS, where the reduced infrastructure footprint saving typically lands at 20% to 30% and the platform absorbs operations work your team should have been doing. Automation matters: Infrastructure-as-Code, GitOps, automated operations, and the SaaS shift onto Microsoft 365, Dynamics and Power Platform routinely cut operating effort by 20% to 35%. Placement alone leaves the other two on the table. The CIOs and CFOs landing the deeper savings use all three together.
Realignment, Not Retreat
This isn't a retreat from cloud. Public cloud continues to carry modernised workloads where elastic scale, pace of innovation and global services earn their keep, and AI workloads sit firmly in that group for a lot of businesess. The difference is that businesses are becoming more intentional about workload placement decisions. Choice, not compromise.
Without a coherent operating model, hybrid complexity becomes a real risk factor. Hybrid itself isn't the problem. It's now the default operational reality. Around three-quarters of UK organisations already run hybrid estates, and most of them got there organically rather than by design. The risk is that different cloud and infrastructure environments accumulate operational nuance, and without consistency across the estate, organisations end up with siloed operating models, duplicated tooling and overheads that reduce capacity to deliver innovation.
I see this pattern every week. Brintons Carpets, a global manufacturer we work with, moved 350 Citrix users to Azure Virtual Desktop in five weeks with zero disruption, and re-platformed core VM-based applications onto UK sovereign infrastructure in parallel. The programme worked because the placement decisions were right and the operating model spanned both legs from week one. One Hybrid Cloud, one FinOps view, one set of dashboards and security governance.
Cloud maturity in 2026 is an orchestration challenge more than a migration exercise. The objective is hybrid infrastructure that operates as a coordinated platform rather than a collection of disconnected technologies. For most organisations, achieving that operational consistency depends on working with partners who can provide cross-platform visibility, governance and expertise without pushing a one-size-fits-all infrastructure model. Without that consistency, complexity itself becomes the source of the inefficiency you were trying to fix.
The successful hybrid strategies of the next three years won't depend on any single platform. They'll depend on whether the operating model around governance, security, cost visibility and workload management is unified across whichever platforms make sense. Get that right and the cloud economics question stops being a board concern. Get it wrong and the bill keeps growing.
